An analyst runs a large aggregation query against a 2 TB table on a Medium warehouse that has been running continuously for the last hour. The exact same query is submitted a second time, but a colleague had just updated a few rows in the table before the second run, so Snowflake does not reuse the result cache. Despite this, the second run of the query completes noticeably faster than the first. Which Snowflake caching mechanism most directly explains the faster second run?